Key predictors of injury severity in occupational accidents involving construction-site vehicles
Across national statistics, construction repeatedly ranks among sectors with the highest injury and fatality rates. Vehicle-related accidents constitute a modest share of minor injuries yet contribute a significant fraction of construction fatalities. This study analysed 16,781 Spanish construction...
| Autores: | , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universidad de Valladolid |
| Repositorio: | UVaDOC. Repositorio Documental de la Universidad de Valladolid |
| OAI Identifier: | oai:uvadoc.uva.es:10324/82506 |
| Acceso en línea: | https://doi.org/10.1016/j.rineng.2025.107762 https://uvadoc.uva.es/handle/10324/82506 |
| Access Level: | acceso abierto |
| Palabra clave: | Ingeniería Industrial Occupational accidents Material agent Construction Vehicles Accident statistics 3310 Tecnología Industrial |
| Sumario: | Across national statistics, construction repeatedly ranks among sectors with the highest injury and fatality rates. Vehicle-related accidents constitute a modest share of minor injuries yet contribute a significant fraction of construction fatalities. This study analysed 16,781 Spanish construction vehicle-related accidents recorded from 2009 to 2022 (2.5% severe-fatal) to identify determinants of injury severity and develop predictive models. Records were retrieved from Delt@, the compulsory national electronic occupational injury reporting platform. Variables were structured into two domains (organisational, contextual) and five categories. Methods combined descriptive profiling, chi 2 association tests, mutual-information ranking and three machine-learning classifiers (Random Forest, XGBoost, multilayer perceptron). Seven predictors-hour block, worker age, job tenure, site zone, deviation pattern, injury type and body region-showed the strongest association with severity. Separate models were trained on contextual and organisational feature sets. The contextual model detected 87.1% of severe/fatal cases (balanced accuracy 88.1.%), while the organisational model detected 59.3% (balanced accuracy 62.1%). The findings emphasise the importance of scheduling (time-of-day exposure), targeted training for short-tenure and at-risk age groups (30-59 years old), and control of the site zone. These results provide practical guidance for managers, regulators, engineers and safety practitioners seeking to reduce the number of vehicle-related accidents on construction sites, particularly those with a high level of severity. |
|---|